1) EXEC: launch the New OpenShift Application wizard and create a new OpenShift application that you import to your workspace. (ex. call the applicatino as ) 2) ASSERT: Your application gets imported to your workspace (ex. project as ) 3) EXEC: delete the new project, kill the OpenShift application 4) EXEC: create the very same application again and have it imported to your workspace 5) EXEC: Go to the Git Perspective and double click some file, so that it gets opened in a editor. You'll very soon get the following error: 6) EXEC: Go to the command-line and do git fsck . Result: You'll spot something very similar to this: [adietish@adietish-thinkpad as2]$ git fsck broken link from commit 6410005677d35ea202df51097c4077f36d8b9ac1 to tree 0a2127cf741ef75aa1d807ad8931366cf63fc579 missing tree 0a2127cf741ef75aa1d807ad8931366cf63fc579 missing blob ef669b78c38fa1317bc89d5d93cd3205c615130f Interesting enough, .gitignore is now a new file in the git repo even though this is completely erroneous. .gitignore was cloned, we only modified it by adding some further entries (.settings, .classpath, .project, etc.): [adietish@adietish-thinkpad as2]$ git status # On branch master # Your branch is ahead of 'origin/master' by 1 commit. # # Changes to be committed: # (use "git reset HEAD <file>..." to unstage) # # new file: .gitignore ALTERNATIVE: 1) EXEC: launch the New OpenShift Application wizard and create a new OpenShift application that you import to your workspace. (ex. call the applicatino as ) 2) ASSERT: Your application gets imported to your workspace (ex. project as ) 3) EXEC: delete the new project 4) EXEC: Reimport the application from OpenShift to your workspace 6) EXEC: Go to the command-line and do git fsck . Result: You'll spot something very similar to this: git fsck broken link from commit 74682dc05a4085a4165ef1a280129e9a9c5d412f to tree 9b742cf75471e6505b5ce7272104f34064326194 missing tree 9b742cf75471e6505b5ce7272104f34064326194 missing blob ef669b78c38fa1317bc89d5d93cd3205c615130f ALTERNATIVE that does not involve using OpenShift: 1) EXEC: create a new java project (ex. test-project) 2) EXEC: share this project with egit: Team->Share Project->Git->Create 3) EXEC: Commit new file (test) to repo: Team->Commit->check ".classpath" 4) EXEC: Delete project in Eclipse, kill Repository-Folder in File-Explorer / CLI 5) EXEC: Repeat steps 1-4 6) Go to CLI and cd to the git repo. Then do: git fsck Result: You'll spot something very similar to this: git fsck broken link from commit 74682dc05a4085a4165ef1a280129e9a9c5d412f to tree 9b742cf75471e6505b5ce7272104f34064326194 missing tree 9b742cf75471e6505b5ce7272104f34064326194 missing blob ef669b78c38fa1317bc89d5d93cd3205c615130f

The OpenShift application that you import into your workspace is shared with a new git repository. If this project is (re)using the name of a git-shared project, that you just deleted, then your new git repository may get corrupted. The root cause is a bug in an EGit repository cache, that's not properly cleaning up when projects get removed. We reported this in https://bugs.eclipse.org/bugs/show_bug.cgi?id=387617 and contributed the fix for EGit 2.1.